How Highly Active K-12 Educators Are Using MagicSchool - Part 2

By JP Martinez Claeys, Valeria Pihl,  Ana T. Ribeiro, Chris Agnew, and Susanna Loeb

Teaching Demands Shaping AI Use: How English Teachers Have Become Power Users

They have access to the same platform, but one teacher runs writing feedback three times, while another builds a worksheet from scratch. They use the same tool for different jobs. 

In our previous post, we analyzed a cohort of approximately 87,000 U.S. educators, representing the top 5% of MagicSchool users. We found that the multi-purpose chatbot is the most used tool in the platform and that its usage represents a similar share of threads for each grade level. Averages conceal substantial variation. The analysis established a distinct grade-level gradient in specific tool use: elementary teachers favor tools supporting communication and student support, while middle and high school teachers trend toward greater use of tools for creating instructional materials. 

To dive deeper into use dynamics, this post focuses on the tools most used by teachers across subject areas. Since most elementary school teachers report teaching “All Core Subjects”, we restrict this analysis to a subsample of middle and high school educators who report teaching one or more specific subjects.1

We focus on the sample of roughly 41,000 power users, educators who logged 52 or more distinct days on MagicSchool since creating their account. Our analysis tracks these teachers’ activity across specific tools, such as Lesson Plan or Writing Feedback, the individual interaction threads where content is generated, and the broader functional categories used to group these workflows. 

Key Findings

  1. Teachers use task-specific tools more frequently than the flexible, multi-purpose chatbot, even though the chatbot could do the same job. The clearest case is writing feedback among English Language Arts (ELA) teachers. ELA teachers can prompt Raina, the multi-purpose assistant, to give feedback on an essay, just as Writing Feedback does, yet these teachers turn to the purpose-built tool far more often. A tool that fits a specific, recurring task, such as providing feedback on student writing, can become embedded in teachers’ workflows.
  2. ELA teachers show a higher level of use of MagicSchool than other subject-specific secondary school teachers. ELA is the most commonly reported subject area on the platform among power users. ELA teachers are unlikely to be more drawn to technology in general than other teachers, so this higher use likely reflects a better fit between the tools and the demands of ELA instruction.
  3. AI tool usage differs across subject areas. ELA teachers are heavier users of Feedback and Assessment tools, largely driven by their use of the Writing Feedback tool. Non-ELA teachers’ primary category of use is Instructional Materials. Non-STEM teachers, whose two most represented subjects are Social Studies and Career and Technical Education, drive this higher usage.

Because ELA is the most commonly reported subject among highly active users, as shown in the subject heatmap from our previous post, we conduct a deeper comparison between ELA and non-ELA teachers. We focus this analysis combining middle and high school educators, because, as established in our previous analysis, teachers in these grade bands exhibit similar platform usage patterns.2

ELA teachers concentrate their AI usage in Writing Feedback, while non-ELA teachers lean more heavily on tools for generating or adapting instructional materials and assessment content.

Figure 1 shows the ten most-used tools among middle and high school educators, which together account for 66.5% of all their tool threads. The clearest differences are that ELA teachers are more than twice as likely to use the Writing Feedback tool as non-ELA teachers are (45% vs 17%) and those who use it come back to it more often. The tool lets teachers describe an assignment, specify a grade level, and get feedback tailored to a rubric or a set of instructions. For ELA teachers, Raina, the multi-purpose chatbot also represents a higher share of their threads compared to non-ELA teachers.

Figure 1: Average Threads per Teacher by Tool, ELA vs. Non-ELA - Top 10 Most Used Tools

Note: Tools shown in bold require a teacher to specify a grade level to generate content. Underneath each tool, a comma-separated list of tool categories is included.

The intensive use of the Writing Feedback tool suggests that tools designed to address specific and important parts of teachers’ instructional practice can provide substantial value. When a tool fits naturally into a recurring task, such as giving feedback on student writing, it may become an important part of a teacher’s workflow and give them a reason to go back to the platform. 

Non-ELA teachers show relatively greater usage of tools such as Text Rewriter, Worksheet Generator, Academic Content, Informational Texts, and Multiple Choice Quiz/Assessment. To break this down further, we split non-ELA teachers into two groups: those who reported teaching STEM/Science or Math ("STEM+"), and all other non-ELA teachers ("Other"). We group STEM/Science and Math together because they involve related quantitative content, and combined they make up the largest subject grouping among non-ELA teachers. Figure 2 repeats the analysis from Figure 1 for these two subgroups.

Figure 2: Average Threads per Teacher by Tool, Non-ELA: STEM+ vs. Other - Top 10 Most Used Tools

We might expect science and math teachers, with problem sets to grade, to be the heaviest users of grading and assessment tools among non-ELA teachers. They are not. STEM+ teachers take fewer actions per tool, on average, than do Other non-ELA teachers, with the exception of Raina. It’s the Other group that drives the higher average number of threads per teacher among non-ELA teachers overall.

These patterns suggest that teachers may gravitate toward AI tools for the types of tasks most common in their subject areas, such as generating instructional materials, creating assessments, or adapting content for classroom use. More broadly, the differences in tool usage suggest that adoption may be shaped by the specific instructional needs and workflows of each discipline.

To understand these patterns, it helps to know who these two groups are. Figure 3 shows how many ELA and non-ELA teachers also reported teaching each of the 10 most common subjects. Because teachers can report more than one subject, these percentages don't need to add up to 100% within each group. ELA teachers most often also teach Special Education (19%) and Social Studies (18%). Non-ELA teachers most often teach Social Studies (30%), Science/STEM (24%), and Career and Technical Education (16%). Combining Science/STEM and Math, roughly 34% of non-ELA teachers report teaching one of these two subjects.

Figure 3: Share of ELA and non-ELA teachers that reported teaching any of the Top 10 most reported subjects.

How Do Functional Workflows Vary Across Subject Areas?

ELA teachers are heavier users of Feedback and Assessment tools. Non-ELA teachers are heavier users of Instructional Materials, and Admin and Communication tools.

We identify the main purposes for which teachers use MagicSchool by grouping each tool-thread activation into categories of use. These categories represent educational or professional domains—such as Instructional Materials, Communication, and Feedback and Assessment—that MagicSchool assigns to each tool. A rubric-based critique of a student essay falls under Feedback and Assessment. A new worksheet on cell division falls under Instructional Materials. An email to families about an upcoming field trip falls under Admin and Communication.

Figure 4: Average Threads per Teacher by Tool Category, ELA vs. Non-ELA - Top 10 Most Used Tools

Figure 4 shows the average number of threads per teacher across tool categories. ELA teachers' two main use categories are Feedback and Assessment and Chatbots, both with higher average number of threads per teacher than non-ELA teachers. 

The gap in the Feedback and Assessment category is largely driven by usage of the Writing Feedback tool. Teachers who incorporate Writing Feedback into their workflow are likely to generate more threads because the tool can support feedback across multiple student submissions or multiple rounds of revision of an assignment. This use pattern differs for other tools in the category, such as the Worksheet Generator or Multiple Choice Quiz/Assessment tools, which teachers may use less frequently or for a single assignment.

The gap in Chatbot use is more surprising. Why would ELA teachers more heavily use a multi-purpose chatbot than do teachers in other subjects? A few candidate explanations are worth considering. ELA instruction may involve needs that MagicSchool’s other tools do not fully address, and Chatbots may simply offer more flexibility for ELA-related tasks than for other subjects. Alternatively, teachers may only be able to incorporate a limited number of specific-purpose tools and the Writing Feedback tool may claim that spot for ELA teachers, leaving them to rely on a multi-purpose tool for everything else. Further analysis could clarify which of these alternatives best explains the pattern.

Non-ELA teachers' two main categories of use are Instructional Materials and Admin and Communication, and in both categories they generate a higher average number of threads per teacher than do ELA teachers. Why would math and science teachers send more emails and administrative messages than English teachers do? As with the tool-level analysis, we split non-ELA teachers into the STEM+ and Other subgroups. Figure 5 shows that the answer isn’t specific to one group. Both STEM+ and Other non-ELA teachers drive the higher usage of Admin and Communication. Despite the common use of problem sets and lab work in STEM+ instructions, STEM+ teachers generate a similar average number of Instructional Materials threads per teacher as ELA teachers. The heavier use comes entirely from the Other subgroup, who average more than 20 threads per teacher, well ahead of both ELA and STEM+ teachers.

Figure 5: Average Threads per Teacher by Tool Category, Non-ELA: STEM+ vs. Other - Top 10 Most Used Tools

One possible explanation: STEM subjects tend to have more standardized readily available curricula, while the Other subgroup, including social studies, career and technical education, special education and others, may have fewer off-the-shelf materials to draw from, pushing those teachers toward tools that build content from scratch. The usage data alone cannot confirm this explanation. More broadly, this pattern suggests that teachers may use AI tools differently depending on the recurring demands of their subject areas: ELA teachers may rely more on tools that support feedback on writing, while non-ELA teachers may rely more on tools that help generate instructional materials. The higher use of Admin and Communication tools among non-ELA teachers is less straightforward to interpret and may reflect differences in workflow, communication needs, or other factors not captured in this analysis.

What Are We Learning?

Three key takeaways emerge from these findings:

  • First, Writing Feedback marks the clearest difference between ELA and non-ELA teachers. More ELA teachers use the tool, and those who use it engage with it more intensively. The higher usage of this tool by ELA teachers suggests that tools designed to support important parts of teachers’ instructional practice can provide substantial value.
  • AI tool usage differs across subject areas, with teachers appearing to use tools that align closely with their classroom needs and instructional workflows. 
  • Overall usage totals can mask differences among educator groups. Examining activity by reported subject provides a fuller picture of which tools and workflows account for activity among highly active users.

These findings show that MagicSchool activity varies not only by grade level, as Part 1 found, but also by reported subject area and the types of tasks supported by different tools.

For the people building these tools, a narrow tool tied to a recurring pain point can out-compete a broad assistant, even one capable of doing the same task. For school and district leaders deciding what to roll out, subject-specific needs may better guide tool selection and teacher training, rather than treating a single tool or training session as sufficient for every teacher.

Looking Ahead

This two-part series offers an initial look at how highly active educators are using AI tools within one platform. Taken together, the findings show that activity on MagicSchool varies by grade level and reported subject. Looking at all users together can hide these differences. 

Future analyses will continue to explore specific teacher workflows, tool-use patterns over time, and emerging differences across diverse educator groups. As artificial intelligence becomes increasingly embedded in schools, tracking how educators actually use these platforms remains an essential step toward understanding the evolving role of technology in K-12 education.

1 For this analysis, a teacher is considered in the ELA group if the user reported ELA explicitly in their subjects and didn’t report All Core Subjects (31.7% of middle and high school teachers), and all remaining teachers are considered in the non-ELA comparison group, with the exception of those that report All Core Subjects, and those with missing subject information (63.6%).

2 For this analysis, a teacher is considered in the ELA group if the user reported ELA explicitly in their subjects and didn’t report All Core Subjects (31.7% of middle and high school teachers), and all remaining teachers are considered in the non-ELA comparison group, with the exception of those that report All Core Subjects, and those with missing subject information (63.6%).